WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: Color Contrast List?


From: James Nurthen
Date: Apr 29, 2013 4:53PM

On Mon, Apr 29, 2013 at 11:31 AM, Jared Smith < <EMAIL REMOVED> > wrote:

> On Mon, Apr 29, 2013 at 11:42 AM, James Nurthen wrote:
> > - Background images under text
> If a user disabled images, it is important that a background color be
> defined that provides adequate contrast for the text. It is my opinion
> that contrast tools should flag contrast issues regardless of whether
> a background image is defined or present.

It is not the false positives due to the defined colour not having adequate
contrast which concern me, but rather issues where there *is* adequate
contrast to the defined colour, but *not* to the background image. I
believe checkers should flag these as warnings which require a manual
(Note - in my experience you can generally eliminate background images
below a certain size which are not repeating from those which have
potential issues)

> At the same time, it would be helpful for contrast tools to also check
> for issues with background images, but this is quite difficult to do
> in an automated way, especially if the image contains both the
> foreground text and the background color.

I agree. This is really hard and I have never seen a good solution.

> > - CSS Gradients
> Yep. This is also quite difficult to do in an automated tool due to
> the many complex ways in which gradients can be defined. There also is
> no specification in WCAG as to how to handle gradients (or text
> highlights, drop shadows, borders, rgba, etc.)

Yep - agree. If anyone wants to write a WCAG technique (or failure) for
this I'm sure they would really appreciate the submission.

> > - Any background CSS colours using rgba values (semi-transparent
> > backgrounds)
> WAVE accounts for semi-transparency though there is no defined WCAG
> formula for doing so. Again, this is fairly complex to handle and we
> need to refine this.

I'd be really interested in seeing your algorithm for this. I've tried
doing something similar but found that when semi-transparent backgrounds
were used they were usually placed ontop of a background image or a
gradient anyway so the previous issues recurred.

> > These techniques are becoming more and more common so these
> semi-automated
> > tools are becoming less and less useful over time.
> Which increases the necessity for common sense and manual testing,
> though I realize the limitations for blind users in manual testing
> these visual components.

Totally agree. Hopefully we can get to a point where the tools reliably
report any certain errors and flag for investigation any places which could
be errors.

> Jared
> > > >